With the general caveat of "YMMV" the drive from the wheel
shaft logic goes like this.
For most sailboats sailing in normal conditions (reasonably balanced and not sailing in gale force conditions) most of the energy consumed by the AP drive is used to overcome the polar moment of inertia of the steering/ rudder system. Small rudder movements usually corresponds to small rudder hydrodynamic forces. If the AP is doing its job rudder corrections are small.
Torques pass through a gearing system as gear
ratio, polar moments of inertia pass through as the square of the gear
ratio. For purposes of discussion assume the gearing ratio(wheel to rudder) is 9 to 1. When you connect directly to the rudder shaft the polar moments of the rudder assy is whatever it is, but backdriving the steering wheel you are driving the polar moment of inertia at a 1 to 81 disadvantage. Now drive from the wheel shaft, the wheel polar moment is what it is, but the rudder assy polar moment is now reduced by the 81 to 1 ratio.
This is why the wheel mounted AP guys (RM wheelpilots, CPT, IPs with thru the legs) all report low energy usage and the rudder shaft drive guys complain about the AHs their AP consumes. The steering shaft drive on my IP 32 was fused at 5 amps (and never blew), now my X5 has current
limit built in so the fuse is no longer needed. When the AP's is driving all I see are little flicks of the ammeter as the course corrections happen. Most of my energy consumption
is to keep the beer
If you can make it work, the steering shaft is the place to apply the force.
"some days you are the dog, and some days the tree"